Much, much more material to cover, so let's get right back into it. I'm loving Alexa Plus, and I think Alexa Plus could end up being in a humanoid form. When can I buy a robot from Amazon?
Well, we've had a consumer robot that we've been offering for a bit, which we started a couple of years ago, but it's still so primitive compared to where it's going to be. I do believe that many, perhaps most consumers will have robots in the home sometime in the next ten years or so.
Ten years. How about three years?
There'll be more. I don't know if it'll be most in three years.
Will they be yours, or am I going to have to get one from Tesla?
I think we're going to keep working on consumer robots. I think you'll see that really great personal assistants like Alexa will be inside these robots, and she'll be able to help you quite a bit inside your home.
All right. Now, July 8th, I intend to engage with Rufus a lot for three days. I'll engage with Rufus. Sometimes Rufus is helpful, and sometimes Rufus should link me to what I want to buy. But Rufus is kind of a dead end at times. When can we have Rufus really know us and know what we like?
Well, I think July 8th you're referencing Prime Day, which actually is four days, 96 hours, double the amount we had last year with millions of deals across every category. I think people are going to be pretty excited, especially at this time, where people are still a little worried about what's going to happen with trade and continuing to have lower prices.
I saw a Reuters piece today that said that already you've had to raise prices from China.
We have, and fortunately we haven't. We did a lot of forward buying several months ago. And then a lot of our sellers are third party selling partners who forward deployed a lot of inventory to avoid some of the issues with the uncertainty around where tariffs are going to settle. And we have so far not seen prices appreciably go up. I think some of that may be the forward buying and deployment I mentioned, but some of it is we have about 2 million sellers in our marketplace. And so even those that decide they're going to pass on whatever the tariff increase is in the form of price, there's probably going to be a number of sellers who decide they're going to take share and not increase prices.
So look, you guys are known, I would think you and an outfit across the street from you, Costco, are the two greatest inflation fighters other than, say, Walmart. But I don't think the government, nor do I think even the American people understand you have your prices. Many cases on Prime Day are going to be like 2018 prices.
When I look at it, very significant discounts. And our stores team, our retail team work very closely with our third party selling partners because we know people are very sensitive about price right now, but we also want to make sure we gave them great deals on items they actually cared about. So I think our customers are going to be very excited July 8th through 11th, and it's going to be a good Prime Day.
Now you have a note in your April letter: 'It does not have to be as expensive as it is today. It won't be in the future.' Chips are the biggest culprit. When I read that, I know you're talking about Nvidia. I know you're developing your own chips, but at the same time, can I make the case that Nvidia actually saves money? That actually does good things for you?
Well, my comment in that letter isn't about any one company. It's just about if you actually want AI to be as expansive as we all believe it can be and should be, we have to decrease the cost, particularly of inference. And the biggest culprit in the cost today happens to be how expensive the chips have been. And so we have a very deep partnership with Nvidia, as a lead partner on every new family of chips they launch. I expect we'll be partners for a long time, but we learned this lesson in the CPU space with our partnership with Intel. Our customers always want better price performance, and so we've learned that we are usually going to have to be the ones that provide that better price performance. And so we have this amazing chip design team in Israel called Annapurna. They helped us build a CPU chip that we call Graviton that we're on the fourth version of. That's 40% more price performance than the leading x86 processors. And 4 or 5 years ago, we saw the same trend happening with silicon and AI, where we set that team a separate team in that organization to go build their own custom silicon AI called Trainium, which we just launched a second version, which is also 30 to 40% more price performance than GPUs. And so we consider it not just an opportunity, but really our responsibility to help customers have lower costs and better price performance, not just on training, but particularly in inference, because the dirty little secret is that while a lot of the expense right now is in training, because everyone's training these models at scale, most of the expense is in inference because you only train periodically, and an application at scale is spitting out inferences all the time.
Okay, now June 17th note: 'Fewer employers not necessarily needed as we roll out generative AI and agents. That should change the way our work is done. We will need fewer people doing some of the jobs that are being done today, and more people doing other types of jobs.' Can you be more specific? And also keep in mind, would it be better to go to a community college or better go to a four year college right now in this world?
